Neil Currie, Crowe’s Chief Information Officer

As Covid entered the world stage, workers and companies solved a problem that many had worked for more than a decade, the ability to make teleworking a reality. Over the night, millions of people were told to work from home, and the infrastructure shifted to support virtual desktops, and meetings via video conferencing became normal. The value of companies like Zoom multiplied, and the beginning of a new operating model appeared. Less noteworthy was the increased introduction of some other new instruments to support a globalized workforce, in which many of the traditional borders had been removed.

These included translation engines, many of which had operated on a smaller scale in the background, but were now becoming part of everyday business. As we know, adoption is driving development, and we are now beginning to see the much-needed sophistication needed to support machine-based translation. This is an area where nuances, context, and ontologies are important. For the sake of a simple example, if you saw the abbreviation “HA” in the text, the context can become very important, for a family doctor it can refer to a headache, a cardiologist for a heart attack and an endocrinologist for hepatitis A. If you saw the word “Body” in the text the business community and its membership association. The translation of the word is affected by the surrounding text and the origin of the document. A manual translation shows several nuances based on a person’s knowledge, experience, and interpretation of what is limited by his or her level of understanding. In the world of machine translation, without these nuances, we see a world where translation is done on the basis of dictionaries tailored to different contexts and environments. Embedded images and non-machine-readable documents in their original format cannot be translated. None of these things come as a surprise to people who use these technologies to support their products, but for the consumer of information, they are still a source of frustration when the recipient expects to receive a complete translation at their disposal. Edible “On-Demand” demands on the evolution of our interactions, the need to fine-tune the level of this secondary background technology is growing and similarly we have seen the industry evolve to provide training materials for machine learning it is likely that we will see such translation dictionaries support this growing need.

We see today that the current flawed model has been introduced in places where accuracy is less important, such as a website that can be translated on request, manuals. In some roles, it is unclear when adoption will become mainstream. These include the lawyer, the profession of general practitioner, and those where a higher level of accuracy is important. In areas such as academic publications, scientific journals, and articles requiring peer review and opinion, they are likely to be translated manually because nuances, interpretation, and environmental differences are required to support the hypothesis proposed and verified by the comparators.

Today, the audio technology of Alexa, Google, and Siri is already a part of our lives, and technologies supported by artificial intelligence that allow them to work in multiple languages ​​in the tones of people in different regions can hopefully be used to support contextual interpretation, nuance. etc. are required for document-based translation to mature as these technologies evolve.



Please enter your comment!
Please enter your name here